bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 FINE-SCALE ADAPTATIONS TO ENVIRONMENTAL VARIATION AND GROWTH 2 STRATEGIES DRIVE PHYLLOSPHERE METHYLOBACTERIUM DIVERSITY. 3 4 Jean-Baptiste Leducq1,2,3, Émilie Seyer-Lamontagne1, Domitille Condrain-Morel2, Geneviève 5 Bourret2, David Sneddon3, James A. Foster3, Christopher J. Marx3, Jack M. Sullivan3, B. Jesse 6 Shapiro1,4 & Steven W. Kembel2 7 8 1 - Université de Montréal 9 2 - Université du Québec à Montréal 10 3 – University of Idaho 11 4 – McGill University 12 13 Key words: Methylobacterium diversity, Phyllosphere community dynamics, rpoB barcoding, 14 temperature adaptation, growth strategies in Bacteria 15 16 Abstract 17 Methylobacterium is a prevalent bacterial genus of the phyllosphere. Despite its ubiquity, little is 18 known about the extent to which its diversity reflects neutral processes like migration and drift, 19 or environmental filtering of life history strategies and adaptations. In two temperate forests, we 20 investigated how phylogenetic diversity within Methylobacterium was structured by 21 biogeography, seasonality, and growth strategies. Using deep, culture-independent barcoded 22 marker gene sequencing coupled with culture-based approaches, we uncovered a previously 23 underestimated diversity of Methylobacterium in the phyllosphere. We cultured very different 24 subsets of Methylobacterium lineages depending upon the temperature of isolation and growth 25 (20 °C or 30 °C), suggesting long-term adaptation to temperature. To a lesser extent than 26 temperature adaptation, Methylobacterium diversity was also structured across large (>100km; 27 between forests) and small geographical scales (<1.2km within forests), among host tree species, 28 and was dynamic over seasons. By measuring growth of 79 isolates at different temperature 29 treatments, we observed contrasting growth performances, with strong lineage- and season- 30 dependent variations in growth strategies. Finally, we documented a progressive replacement of 31 lineages with a high-yield growth strategy typical of cooperative, structured communities, in

1 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

32 favor of those characterized by rapid growth, resulting in convergence and homogenization of 33 community structure at the end of the growing season. Together our results show how 34 Methylobacterium is phylogenetically structured into lineages with distinct growth strategies, 35 which helps explain their differential abundance across regions, host tree species, and time. This 36 works paves the way for further investigation of adaptive strategies and traits within a ubiquitous 37 phyllosphere genus. 38 39 Abstract importance 40 Methylobacterium is a bacterial group tied to plants. Despite its ubiquity and importance to their 41 hosts, little is known about the processes driving Methylobacterium community dynamics. By 42 combining traditional culture-dependent and –independent (metagenomics) approaches, we 43 monitored Methylobacterium diversity in two temperate forests over a growing season. On the 44 surface of tree leaves, we discovered remarkably diverse and dynamic Methylobacterium 45 communities over short temporal (from June to October) and spatial scales (within 1.2 km). 46 Because we cultured very different subsets of Methylobacterium diversity depending on the 47 temperature of incubation, we suspected that these dynamics partly reflected climatic adaptation. 48 By culturing strains in lab conditions mimicking seasonal variations, we found that diversity and 49 environmental variations were indeed good predictors of Methylobacterium growth 50 performances. Our findings suggest that Methylobacterium community dynamics at the surface of 51 tree leaves results from the succession of strains with contrasted growth strategies in response to 52 environmental variations. 53 54 Acknowledgments: FRQNT (funding), NSERC (funding), Canada Research Chairs (funding), 55 NSF (DEB-1831838), Geneviève Lajoie, Dominique Tardif, Ariane Lafrenière, Hélène Dion- 56 Phénix, Yves Terrat, Kenta Araya (help for sampling), Sylvain Dallaire (help for phenotyping 57 screen), Sergey Stolyar (discussions), Gault Natural Reserve (McGill University), Station 58 Biologique des Laurentides (UdeM), Centre d’Étude de la Forêt (CEF), QCBS. 59 60 Author contributions : J.B.L., E.S.L., D.C.M., G.B., B.J.S. and S.W.K. planned fieldwork and 61 the experiments. J.B.L. E.S.L., D.C.M. and G.B. performed experiments. J.B.L., E.S.L., D.S. and 62 J.M.S performed bioinformatic analyses. J.A.F., C.J.M. and J.M.S. provided discussion in the

2 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

63 early stages of this study. J.B.L. B.J.S. and S.W.K. drafted the manuscript with contributions 64 from C.J.M. 65 66 Data accessibility: raw reads for 16s and rpoB barcoding on phyllosphere communities 67 (BioProject PRJNA729807; BioSamples SAMN19164946-SAMN19165146) were deposited in 68 NCBI under SRA Accession Numbers SRR14532212-SRR14532451. Partial nucleotide 69 sequences from marker genes obtained by SANGER sequencing on Methylobacterium isolates 70 (BioProject PRJNA730554 ; Biosamples SAMN19190155-SAMN19190401) were deposited in 71 NCBI under GenBank Accession Numbers MZ268514-MZ268593 (16s), MZ330152-MZ330358 72 (rpoB) and MZ330130-MZ330151 (sucA). R codes and related data were deposited on Github 73 (https://github.com/JBLED/methylo-phyllo-diversity). 74

75

3 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

76 Introduction 77 78 Phyllosphere, the aerial parts of plants including leaves, is a microbial habitat estimated as vaste 79 as twice the surface of the earth (1). Although exposed to harsh conditions, like UVs, temperature 80 variations and poor nutrient availability, the phyllosphere arbors a diverse community of 81 microorganisms, of which bacteria are the most abundant (1). A key challenge in microbial 82 ecology and evolution is understanding the evolutionary and ecological processes that maintain 83 diversity in habitats such as the phyllosphere. Bacteria living in the phyllosphere carry out key 84 functions including nitrogen fixation, growth stimulation and protection against pathogens (1–3). 85 At broad spatial and temporal scales, bacterial diversity in the phyllosphere varies as a function 86 of biogeography and host plant species, potentially due to restricted migration and local 87 adaptation to the biotic and abiotic environment (4–6), leading to patterns of cophylogenetic 88 evolutionary association between phyllosphere bacteria and their host plants (7). Whether those 89 eco-evolutionary processes are important at short time scales, as microbes and their host plants 90 migrate and adapt to changing climates, is still an open question (8). Another challenge is to link 91 seasonal variation with plant-associated microbial community dynamics, as shifts in microbial 92 community composition are tighly linked with host plant carbon cycling (9) and ecosystem 93 functions including nitrogen fixation (10). More generally, we understand very little about how 94 the ecological strategies of phyllosphere bacteria vary among lineages and in response to 95 variation in environmental conditions throughout the growing season (9, 11). 96 97 Phylogenetic signal in traits or suites of correlated traits that are used as indicators of the 98 ecological life history strategy of microbes suggest niche adaptation (12), and these phenotypic 99 traits influence the assembly of ecological communities through their mediation of organismal 100 interactions with the abiotic and biotic environment (13). Recent work evaluating the 101 phylogenetic depth of microbial trait evolution has shown that many microbial phenotype traits 102 exhibit phylogenetic signal, with closely related lineages possessing similar phenotypic traits, 103 although the phylogenetic depth at which this signal is evident differs among traits (14). Most 104 comparative studies of microbial trait evolution have focused on broad scale patterns across 105 major phyla and classes (14), although some studies have found evidence for complex patterns of 106 phenotypic and genomic evolution within microbial genera interacting to determine patterns of

4 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

107 community assembly and niche evolution (15, 16). Furthermore, to date the majority of studies of 108 the diversity of plant-associated microbes have been based on the use of universal marker genes 109 such as the bacterial 16S rRNA gene, providing a global picture of long-term bacterial adaptation 110 to different biomes and host plants at broad phylogenetic scales (17), but lacking sufficient 111 resolution to assess the evolutionary processes at finer spatial and temporal scales that lead to the 112 origin of and adaptations within microbial genera and species (18, 19). 113 114 The Rhizobiales genus Methylobacterium (Alphaproteobacteria, Rhizobiales, 115 Methylobacteriaceae) is one of the most prevalent bacterial genera of the phyllosphere, present 116 on nearly every plant (20, 21). Characterized by pink colonies due to carotenoid production, 117 Methylobacterium are facultative methylotrophs, able to use one-carbon compounds, such as 118 methanol excreted by plants, as sole carbon sources (21–23). Experimental studies have shown 119 the important roles of Methylobacterium in plant physiology, including growth stimulation 120 through hormone secretion (24–26), heavy metal sequestration (26), anti-phytopathogenic 121 compound secretion, and nitrogen fixation in plant nodules (27), sparking increasing interest in 122 the use of Methylobacterium in plant biotechnology applications (26, 28, 29). Although up to 64 123 Methylobacterium species have been described (30–38), genomic and phenotypic information 124 was until recently limited to a small number of model species: M. extorquens, M. populi, M. 125 nodulans, M. aquaticum and M radiotolerans, mostly isolated from anthropogenic environments, 126 and only rarely from plants (39–43). Newly available genomic and metagenomic data now allow 127 a better understanding of Methylobacterium diversity across biomes (30), but we still understand 128 relatively little about the drivers of the evolution and adaptation of Methylobacterium in natural 129 habitats. 130 131 In this study, we assessed the diversity of Methylobacterium in temperate forests and asked 132 whether Methylobacterium associated with tree leaves act as an unstructured population with 133 large effective size, or if this diversity was maintained by regional factors (e.g. a combination of 134 isolation by distance and regional environmental variation) or by niche adaptation (e.g. host tree 135 or temperature adaptation) (12). First, we assessed Methylobacterium diversity by combining 136 culturing and barcoding approaches along with phylogenetic analysis and quantified how this 137 diversity varied across space, time, and environment in the phyllosphere. Second, we quantified

5 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

138 the extent of phylogenetic niche differentiation within the bacterial genus Methylobacterium, 139 with a focus on quantifying the evidence for adaptation to local environmental variation at 140 different spatial, temporal and phylogenetic scales. We hypothesized that distinct phylogenetic 141 lineages would be associated with distinct environmental niches. Third, we quantified 142 Methylobacterium growth performance under fine-scale environmental variations, with a focus 143 on temperature, to determine whether Methylobacterium diversity fine-scale dynamics over space 144 and time might result from environmental filtering of isolates with contrasting growth strategies 145 under local environmental conditions. We found that Methylobacterium phyllosphere diversity 146 consisted of deeply branching phylogenetic lineages associated with distinct growth phenotypes, 147 isolation temperatures, and large-scale spatial effects (forest of origin), while finer-scale spatial 148 effects, host tree species, and time of sampling were more weakly and shallowly phylogenetically 149 structured. Over the course of a year, from spring to fall, we observed a homogenization of 150 Methylobacterium community structure coinciding with the progressive replacement of isolates 151 with high yield strategy by isolates with rapid growth. Together our results show that this 152 ubiquitous phyllosphere genus is structured into lineages with distinct growth strategies, which 153 helps explain their differential abundance across space and time. 154 155 Results 156 157 Phylogenetics of plant-associated Methylobacterium diversity. 158 159 We evaluated the known Methylobacterium diversity associated with plants, especially the 160 phyllosphere, by compiling information about the origin of 153 Methylobacterium isolates for 161 which genomes were available. We found that plants (65% of genomes) and especially the 162 phyllosphere compartment (41% of genomes) were the most prevalent source of 163 Methylobacterium. From these genomes, we built a phylogenetic tree based on the complete 164 nucleotide sequence of rpoB, a highly polymorphic marker that experienced no copy number 165 variation in many bacteria taxa (44, 45), and that we confirmed to be single copy in 166 Methylobacterium and related genera Microvirga and Enterovirga (Figure 1, Supplementary 167 dataset 1a). In this phylogeny, we roughly identified main Methylobacterium groups (A, B, and 168 C) previously defined based on the 16S gene (30) and found that phyllosphere-associated

6 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

169 diversity was not randomly distributed in the Methylobacterium phylogenetic tree. Isolates from 170 the phyllosphere represented the largest part of diversity within group A (56% of isolates) but not 171 in groups B and C (17 and 12% of isolates, respectively). Most of diversity within group A 172 consisted of undescribed taxa falling outside previously well-described linages (Figure 1, 173 Supplementary dataset 1a). For the purpose of this study, we refined Methylobacterium group 174 A that we subdivided into 9 monophyletic clades (A1-A9), using a ∼92% pairwise similarity (PS) 175 cut-off on the rpoB complete sequence. 176 177 16S Community analyses reveal Methylobacterium ubiquity and diversity in the phyllosphere. 178 179 We focused on Methylobacterium phyllosphere diversity variation observable at the scale of 180 seasonal variation on individual trees within a geographic region, the temperate forests of 181 northeastern North America (Figure 2a,b). In two forests: Mont Saint Hilaire (MSH; 45.54 N 182 73.16 W ; Figure 2c) and Station biologique des Laurentides (SBL; 45.99 N 73.99 W ; Figure 183 2d), we marked 40 trees representative of diversity observed in 4-6 plots distributed along a 1.2 184 km transect (4-10 trees per plot). In MSH, the transect followed an elevation and floristic 185 gradient, while in SBL, it followed a relatively constant environment. For this time series, each 186 tree was sampled 3-4 times from June to October 2018 (Figure 2b; Supplementary dataset 1b). 187 We evaluated the microbial phyllosphere diversity in our time series based on sequencing and 188 identification of bacterial 16S gene amplicon sequence variants (ASVs; (46)) in a representative 189 subset of 46 phyllosphere samples from 13 trees (Supplementary dataset 1c-e). As observed in 190 previous studies (4), the distribution of the phyllosphere bacterial community was mostly 191 explained by differences among forests (31.6% of variation explained; p<0.001; PERMANOVA; 192 Hellinger transformation; 10,000 permutations), host tree species (15.6% of variation; p<0.001) 193 and time of sampling (12.0%; p<0.05; Table 1), indicating that between-forest variation and 194 adaptation to hosts were the main drivers of this diversity, which also varied greatly at the time 195 scale of a year. Although representing only 1.3% (0.0-3.2% per sample) of total 16S sequence 196 diversity, Methylobacterium was present in almost all analyzed samples (45 out of 46; 197 Supplementary dataset 1d,e). We assigned the 15 Methylobacterium ASVs identified by 16S 198 sequencing to clades from Methylobacterium group A: A9 (M. phyllosphaerae/M. 199 mesophilicum/M. phyllostachyos/ M. pseudosasicola/M. organophilum; 0.87% of total diversity,

7 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

200 nine ASVs), A6 (M. sp., 0.29%; one ASV) and A1 (M. gossipicola; 0.13%, 3 ASVs; 201 Supplementary dataset 1e). With two rare ASVs (<0.01% of relative abundance) assigned to M. 202 komagatae, belonging to group A (30) but unrelated to any aforementioned clade, we defined a 203 new clade (A10). No ASV was assigned to group B or group C, hence confirming observations 204 from available genomes that Methylobacterium group A is tightly associated with the 205 phyllosphere. 206 207 Development of a single-copy molecular marker to monitor fine-scale dynamics of 208 Methylobacterium populations. 209 210 Although the 16S barcoding approach suggests that Methylobacterium is ubiquitous in the 211 phyllosphere of temperate forests, regardless of location, time and host tree species, identification 212 based on 16S sequencing presents some limits in assessing microbial population dynamics at 213 local scales, and in assessing fine-scale evolutionary adaptations (18). First, the low 214 polymorphism of the 16S rRNA gene does not permit distinguishing among species within clades 215 typical of the phyllosphere, thus confounding species with potentially divergent evolutionary 216 routes reflecting the distinct ecological niches they occupy within the phyllosphere. Second, 16S 217 copy number variation within Methylobacterium (4-12) and even within groups of closely related 218 isolates (4-6 copies in group A; 5 in group B; 6-12 copies in group C), may induce biases in 219 estimating relative abundances of taxa, and thus in assessing the dynamics of populations over 220 space, time and among host tree species. 221 222 To assess Methylobacterium diversity at a finer evolutionary level, we thus developed a 223 molecular marker targeting all members of the Methylobacteriaceae family, using the core gene 224 rpoB (Figure 1 (44, 45)). Based upon rpoB sequences available for Methylobacterium 225 (Supplementary dataset 1a), as well as sequencing of rpoB partial sequences from 20 226 Methylobacterium isolates from a pilot survey in MSH in 2017 (Figure 2d; Table S1; 227 Supplementary dataset 1f,g) we determined that this gene is polymorphic enough to explore 228 diversity within the aforementioned Methylobacterium clades (See supplementary method). For 229 the rest of this study, we used the rpoB marker to monitor temporal trends in Methylobacterium 230 diversity in the phyllosphere from our 2018 time series in both forests.

8 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

231 232 Culture-based assessment of Methylobacterium diversity in the tree phyllosphere. 233 234 We evaluated the culturable part of Methylobacterium diversity from a subsample of 36 trees (18 235 per forest) representative of floral diversity in the 2018 time series in MSH and SBL. To date, 236 Methylobacterium was mostly isolated assuming its optimal growth was in the range 25-30 °C 237 (47), an approach that could lead to a bias toward mesophylic isolates in estimating microbial 238 diversity, especially in the case of microbes inhabiting temperate forests where temperatures 239 typically range from 10 to 20 °C during the growing season (48). We thus performed replicate 240 isolation of Methylobacterium at both 20 and 30 °C on minimum mineral salt (MMS) media with 241 0.1% methanol as sole carbon source. We successfully amplified the rpoB marker for 167 pink 242 isolates that we assigned to Methylobacterium based upon their phylogenetic placement 243 (Supplementary dataset 1g,h; Supplementary method). As observed for 16S ASVs, most 244 isolates were assigned to clades from group A typical of the phyllosphere: A9 (59.9% of isolates), 245 A6 (24.6%), A1 (5.4%), A10 (3.6%) and A2 (1.8%). Few isolates were assigned to group B 246 (4.2% of isolates), mostly related to M. extorquens, and none to group C (Table S2). But the 247 higher polymorphism in the rpoB marker allowed us to uncover a considerable diversity within 248 clades, as we identified 71 unique rpoB sequences, in contrast to the smaller number obtained 249 with 16S barcoding (15 ASVs). 250 251 Such high diversity in rpoB suggests that standing genetic variation is segregating within 252 Methylobacterium populations inhabiting the phyllosphere. We hypothesized that the 253 maintenance of this divestity could be explained by regional factors (e.g. a combination of 254 isolation by distance and regional environmental variation) and/or by niche adaptation (e.g. host 255 tree or temperature adaptation)(12). To do so, we quantified associations between 256 Methylobacterium diversity assessed at varying depths in the rpoB phylogeny (Supplementary 257 method; Figure 3a) with four factors and their interactions (PERMANOVA with 10,000 258 permutations): (1) forest of sampling (2) temperature of isolation, (3) sampling time, and (4) host 259 tree species. For every phylogenetic depth tested, diversity had distinct associations with forest of 260 origin (4.5±1.0% of variance explained; p<0.001) and temperature of isolation (5.9±2.1% of 261 variance explained; p<0.001; Figure 3a; Supplementary dataset 1i). Interestingly, temperature

9 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

262 of isolation was the most important factor distinguishing deep phylogenetic divergences (pairwise 263 nucleotide similarity range: 0.948-0.993), while forest of origin was slightly more important in 264 structuring more recently diverged nodes (pairwise nucleotide similarity >0.993). This suggests 265 that mechanisms underlying isolation success at different temperatures played a significant role 266 in the early divergence among Methylobacterium clades associated with tree leaves, while spatial 267 variation among different forests evolved more recently. Time of sampling had a slight but 268 significant effect on diversity (2.1±0.2% of variance explained; p<0.05) and it was only observed 269 for higher pairwise nucleotide similarity values (range 0.994-1.000), suggesting seasonal 270 Methylobacterium diversity dynamics at the tip of the tree. We did not observe any significant 271 effect of host tree species on Methylobacterium isolate diversity, for any level of the phylogeny, 272 suggesting no strong specific associations between Methylobacterium isolates and the tree species 273 from which they were isolated. 274 275 We next asked specifically which nodes within the Methylobacterium phylogenetic tree were 276 associated with the two major factors contributing to overall diversity, namely forest and 277 temperature of isolation (Figure 3a). For every level in the rpoB phylogeny, we independently 278 tested for nodes (with at least 30% of support) associated with forest of origin (SBL and MSH) or 279 temperature of isolation (20 and 30 °C) by permutation (100,000 permutations per level and per 280 factor; Figure 3b). We identified two nodes strongly associated with temperature of isolation and 281 corresponding to clades A6 (20 °C; p<0.001) and A9+A10 (30 °C; p<0.001; Figure 3b). Other 282 clades were evenly isolated at 20 and 30 °C and we observed no significant association between 283 temperature of isolation and nodes embedded within clades. Nodes associated with forest of 284 origin also roughly corresponded to certain major clades, with clades A1+A2 almost exclusively 285 sampled in MSH (p<0.01). Overall, clade A9 was isolated significantly more often in SBL 286 (p<0.001) but at least three of its subclades were significantly associated to either MSH or SBL 287 (p<0.05), suggesting relatively recent migration events. The fact that Methylobacterium diversity 288 typical of the phyllosphere show contrasted associations with forest and temperature of isolation 289 suggests that their evolution was tightly linked with processes related to spatial and 290 environmental variation including isolation by distance as well as niche-based processes 291 including adaptation to local climatic conditions. 292

10 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

293 Comparison of Methylobacterium diversity assessed by rpoB barcoding and isolation 294 295 To determine if these culture-based results were representative of the potentially uncultured 296 Methylobacterium diversity, we developed a culture-independent barcoded amplicon sequencing 297 approach based on rpoB. We performed rpoB amplicon sequencing for 179 leaf samples from 53 298 trees in both forests, allowed a monthly monitoring for most trees (Supplementary dataset 1c,j). 299 We identified 283 Methylobacteriaceae rpoB ASVs in these samples (Supplementary dataset 300 1j,k), representing 24.6% of all sequences. Non-Methylobacteriaceae ASVs were mostly 301 assigned to other Rhizobiales families (850 ASVs, 70.33% of sequence abundance) and to 302 Caulobacterales (209 ASVs, 4.42% of sequence abundance) typical of the phyllosphere (see 303 Supplementary method), suggesting that our rpoB marker is not limited to Methylobacteriaceae 304 and can potentially be used at a broader taxonomic scale (Figure S1a). Within 305 Methylobacteriaceae, ASVs were mostly classified as Methylobacterium (200 ASVs, 23.05% of 306 sequence relative abundance), and Enterovirga (78 ASVs, 1.56%; Supplementary dataset 1k). 307 We assigned Methylobacterium ASVs to previously defined clades using a maximum likelihood 308 tree combining ASV sequences and reference genomes (Figure S1b). Most of Methylobacterium 309 diversity was within the previously cultured clades A9 (45.2% of Methylobacterium sequence 310 abundance), A6 (24.3%), A1 (6.1%) and A10 (1.0%; Supplementary dataset 1j; Table S2). 311 Estimates of Methylobacterium diversity based on rpoB sequences from culture-independent 312 sequencing or cultured isolates were generally concordant (Figures S1c,d; Table S2). 313 Nevertheless, we cannot exclude the possibility that some diversity was not isolated. For 314 instance, although 19.1% of total Methylobacterium diversity assessed by rpoB culture-free 315 barcoding was assigned to group B, it only represented 4.2% of isolates. One possible 316 explanation could be adaptation of some isolates from this group to temperatures below the range 317 used for isolation (20-30 °C), as temperatures at the very beginning (May) and end (October) of 318 the growing period in Quebec typically range between 5 and 15°C. 319 320 Fine-scale temporal and spatial distribution of Methylobacterium diversity assessed by rpoB 321 barcoding 322

11 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

323 Our isolate-based survey suggested that the distribution of Methylobacterium diversity was 324 driven by both spatial and seasonal variation. This result was largely supported using the culture- 325 free approach (rpoB barcoding). Specifically, we found that spatial variation at both large 326 (distance between forests: 100km) and local scales (distance between plots within forest: 150- 327 1,200 m), as well as sampling date during the growing season (1-5 months), explained the largest 328 part of variance in the community composition of 200 Methylobacterium ASVs (proportion of 329 variation explained: 32.4%, 8.0% and 4.8%, respectively; p<0.001; PERMANOVA; Hellinger 330 transformation, Bray-Curtis dissimilarity, 10,000 permutations; Table 1). A large proportion of 331 Methylobacterium ASVs (83 out of 200) were significantly associated with one or either forest 332 (ANOVA; Bonferroni correction; Figure 4a; Supplementary dataset 1l), regardless their clade 333 membership. The only exception was observed for clade A1, which was almost exclusively 334 observed (and isolated; see Figure 3b) in the MSH forest. Also consistent without our isolation- 335 based results, we found no clear association between ASV or clade with host tree species, nor 336 plots within forests (data not shown). 337 338 To focus on temporal variation and fine-scale spatial effects, we removed large-scale spatial 339 variation by analyzing each forest separately. We quantified fine-scale spatial and temporal 340 dynamics of Methylobacterium diversity (200 ASVs), using autocorrelation analysis based on 341 Bray-Curtis dissimilarity index (BC). We observed a weak but significant decrease of community 342 similarity with geographical distance separating two samples within MSH (ANOVA on linear 343 model; p<0.001) but not SBL (p>0.05, Table 2, Figure 4b), and a significant decrease of 344 community similarity with time separating two samples in both forests (ANOVA on linear 345 model; p<0.001; Table 2), which was more marked in MSH than in SBL (Figure 4c). Both 346 results indicate that Methylobacteriaceae diversity is heterogeneously distributed even at very 347 local space and time scales. The overall community dissimilarity consistently decreased from 348 June to October in both MSH (from 0.624 to 0.297) and SBL (from 0.687 to 0.522; Table 2, 349 Figure 4d), suggesting that Methylobacteriaceae diversity was progressively homogenized by 350 migration or ecological filtering between the beginning and the end of the growing season at the 351 scale of a forest, although without affecting locally its heterogeneous spatial distribution in MSH 352 (Table 2, Figure 4e). The heterogeneous spatial distribution of Methylobacterium diversity 353 observed in MSH forest suggests either more restricted migration than in SBL, and/or local

12 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

354 adaptation due to environmental gradients. One possible explanation is that the floristic and 355 altitudinal gradient observed in MSH (Figure 2d) – but not in SBL – might generate local 356 adaptation to host tree species for Methylobacterium colonizing leaves in this more 357 heterogeneous forest, hence counteracting the homogenizing effect of neutral processes like 358 migration (49). Accordingly, we found slight but significant effects of host tree species, and of 359 the interaction between host tree species and plots within forests, on Methylobacterium 360 community composition (explaining 7.1% and 4.3% of variation in community composition; 361 p<0.001 and ,p<0.01, respectively; PERMANOVA; Table 1). 362 363 We tested for temporal autocorrelation in each node of the ML tree (Figure S1b) supported by at 364 least 30% of bootstrapps (200 permutations) and observed significant temporal dynamics (as 365 attested by the positive slope between pairwise time and BC dissimilarity) in most testable nodes 366 (ANOVA, Bonferroni correction; Figure 4f). We observed the strongest temporal signals in 367 nodes embedded within clades A1 (MSH) and B (both forests), Accordingly, we found 25 ASVs 368 whose abundance significantly increased thourough the growing season (ANOVA; Bonferroni 369 correction; p<0.05), mostly belonging to clades A1 (n=11). Four ASVs increased significantly 370 with time in both forests and mostly belonged to group B (n=3), suggesting that at least a part of 371 the characteristics driving short-term dynamics of Methylobacterium communities are shared 372 among members of different clades within the genus (Supplementary dataset 1l). 373 374 Effect of short scale temperature variation in combination with other environmental and genetic 375 factors on Methylobacterium growth performances 376 377 A major environmental difference between forest sites and throughout the growing season relates 378 to shifts in temperature, suggesting, together with temperature preference of some clades during 379 the isolation step, that Methylobacterium diversity dynamics we observed over space and time 380 might result from clade contrasted growth performances under local climatic conditions. To 381 further explore the role of temperature, we measured growth of 79 Methylobacterium isolates 382 (sampled in 2018 in both forests; MSH: n=32, SBL: n=47 ; Supplementary dataset 1m) for four 383 temperature treatments mimicking temperature variations during the growing season. Each 384 treatment consisted of an initial pre-conditioning step (P) during which each isolate was

13 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

385 incubated on solid MMS media with methanol as sole carbon source for 20 days at either 20 °C 386 (P20) or 30 °C (P30), and a second monitoring step (M) during which pre-conditioned isolates 387 were incubated on the same media and their growth monitored for 24 days at 20 °C (P20M20 and 388 P20M30) or 30 °C (P30M20 and P30M30; Figure S2). Treatments P20M20 and P30M30 389 mimicked stable thermal environments, and treatments P20M30 and P30M20 mimicked variable 390 thermal environments. For each isolate and temperature treatment, logistic growth curves were 391 inferred from bacteria spot intensity variation observed over three time points during the 392 monitoring step (Figures S2, S3). From growth curves, we estimated maximum growth intensity, 393 or yield (Y) and growth rate (r) as the inverse of lag+log time necessary to reach Y (Figure S4 394 (50, 51)). Clade membership explained a large part of variation in Y and r (30.6 and 7.6% of 395 variation explained, respectively; ANOVA; p<0.001), indicating that Methylobacterium growth 396 performance is largely long-term inherited and tends to be shared among clade members (Figures 397 5a,b, Table 3). Certain clades had a higher yield range than others, suggesting differences in their 398 carbon use efficiency (51), here provided by methanol. For example, group B isolates (Y = 12.2 ± 399 5.0) have higher yield than group A (Y = 5.4 ± 3.5). Different growth rates rather suggest 400 contrasted growth strategies across clades (51, 52). Isolates from clades A1, A2 and B had the 401 highest growth rate (r range: 0.101±0.032 – 0.121±0.031), suggesting they have fast-growth 402 strategy. Other clades (A6, A9 and A10) had on average slower growth (r range: 0.082±0.021 – 403 0.088±0.024), suggesting that they have more efficient strategy (51). 404 405 Compared to clade membership, we observed that time of sampling, host tree species and forest 406 explained less variation in growth rate (5.4%, p<0.001; 2.2%, p<0.01 and 1.5%, p<0.05, 407 respectively; ANOVA; Table 3), suggesting that plasticity to the environment was a secondary 408 but still determining factor in strain growth strategy. The weak or non-significant interactions 409 between clade membership and the aforementioned environmental factors (ANOVA; Table 3) 410 suggest that these patterns were consistent across clades, indicating they are unlikely to result 411 from long-term adaptation but rather correspond to short-term responses to environmental 412 conditions. In both SBL and MSH, growth rate increased consistently from June (r = 413 0.075±0.018 and 0.085±0.033, respectively) to September/October (r = 0.097±0.031 and 414 0.103±0.027, respectively; Figure 5c). Growth rate increase thorough the season might result 415 from increasing competition within phyllosphere communities, for instance by selection for faster

14 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

416 growth. Accordingly, culture-independent rpoB confirmed that clades associated with faster 417 growth (clades B and A1) increased in abundance over time. 418 419 Temperature also had significant effects on growth performance. Temperature during the 420 monitoring phase explained respectively 2.0% and 15.8% of variation in yield and growth rate 421 (p<0.01 and p<0.001, respectively; ANOVA; Figure 5d, Table 3), regardless of clade 422 membership (no significant interaction in the ANOVA). Isolates incubated at 20 °C have on 423 average higher yield (Y=6.9±5.4) but slower growth (r=0.077±0.022) than isolates incubated at 424 30 °C (Y=4.9±3.6; r=0.100±0.030), suggesting that higher temperature tends to shift 425 Methylobacterium toward faster growth but lower efficiency, a typical trade-off observed in other 426 bacteria (51). The effect of monitoring temperature on growth rate was also independent from 427 time of sampling (no significant interaction in the ANOVA), suggesting that Methylobacterium 428 also grow faster and less efficiently at 30 °C regardless temporal environmental variations. 429 Accordingly, the pre-conditioning temperature had no effect on growth rate (p>0.05; ANOVA), 430 and very limited on yield (1.4%; p<0.05; ANOVA), suggesting limited effect of short-time 431 temperature shift on Methylobacterium growth performance (Table 3). 432 433 Discussion 434 435 Methylobacterium is ubiquitous on leaves in the temperate forests of Québec and its diversity in 436 this habitat is quite similar to what has been described in the phyllosphere throughout the world, 437 with three main clades A9 (M. brachiatum, M. pseudosasicola), A6 (M. sp.) and A1 (M. 438 gossipicola) dominating diversity in the canopy. Our barcoding approach based on a clade- 439 specific rpoB marker revealed astonishing diversity within these clades, as well as within several 440 other clades: B (M. extorquens), A2 (M. sp.), A4 (M. gnaphalii, M. brachytecii) and A10 (M. 441 komagatae ) whose importance in the phyllosphere has been underestimated by classical 16S 442 barcoding or isolation approaches. This diversity, like that of the overall phyllosphere 443 community, was mostly determined by differences between forests, with barcoding approaches 444 suggesting combined effects of restricted migration, local adaptation to host tree species, and 445 climatic conditions at large geographical scales (>100km). With higher molecular resolution, we 446 observed that Methylobacterium diversity was structured even at the scale of a forest (within 2

15 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

447 km), which according to a fine-scale timeline survey over the growing season also showed a clear 448 pattern of temporal dynamics and succesion. A finer analysis of Methylobacterium diversity 449 suggested that clade identity partly explained Methylobacterium geographical distribution at large 450 scale (forest) but not at finer scales (plots), nor was it an indicator of adaptation to a particular 451 host tree species, nor a determinant of temporal dynamics. Rather, the distribution of 452 Methylobacterium diversity at small temporal and geographical scales likely resulted from more 453 contemporaneous community assembly events selecting for phenotypic traits that evolved among 454 deeply diverging lineages of Methylobacterium, as has been observed in other bacterial (16) and 455 plant clades (53). 456 457 We explored mechanisms explaining the temporal dynamics of Methylobacterium diversity at the 458 scale of a growing season. Because we observed contrasting Methylobacterium isolable diversity 459 between 20 and 30 °C, we suspected that adaptation to temperature variation during the growing 460 season could explain part of these temporal dynamics. By monitoring Methylobacterium isolate 461 growth under different temperature treatments, we confirmed that temperature affected isolate 462 growth performances. The fact that most tested isolates grow slower but more efficiently at 20 °C 463 than at 30 °C (Figure 5d), regardless of their phylogenetic and environmental characteristics, is 464 in line with a temperature-dependent trade-off between growth rate and yield described in many 465 bacteria (reviewed in (51)). High yield strategies are typical of cooperative bacterial populations, 466 while fast growth-strategies are typical of competitive populations (51), suggesting that 20 °C is 467 likely closer to the thermal niche of a cooperative Methylobacterium community, in agreement 468 with average temperatures in temperate forests during the growing season. This observation also 469 stresses the importance of considering incubation temperature when interpreting results from 470 previous studies assessing Methylobacterium diversity based on isolation. We observed several 471 lines of evidence that factors other than direct adaptation to temperature drive Methylobacterium 472 responses to temperature variation, by affecting their growth strategy in different competitive 473 conditions rather than by affecting their metabolism directly. First, clade identity was one of the 474 main predictors of overall isolate performance, with some clades (A1, A2, B) possessing a rapid 475 growth strategy under all temperature conditions, while others (clades A6, A9, A10) had 476 systematically slower growth. These clade-specific growth strategies could explain for instance 477 why certain Methylobacterium isolates are less competitive and less frequently isolated at higher

16 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

478 temperatures. Still, we cannot rule out that clade-specific growth strategy also reflect 479 experimental conditions. Second, we observed strong associations between isolates growth 480 performance and time of sampling, regardless of clade association, suggesting that growth 481 dynamic strategies also respond to seasonal variations in environmental conditions, and to the 482 level of establishment and competition in the phyllosphere community (51). This observation, 483 together with clade identity, could explain why, assuming that environmental conditions at the 484 end of the growing season became unfavorable for most strains, isolates from clades A1 and B 485 with a fast-growth strategy consistently increase in frequency during this period and lead to the 486 homogeneization of the community. Taken together, our temporal survey of diversity dynamics 487 and screening for growth performance suggest the following timeline of the dynamics of the 488 Methylobacterium phyllosphere community. At the very beginning of the growing season, a pool 489 of bacteria with mixed ecological strategies and genotypes colonizes newly emerging leaves. Due 490 to the stochasticity of this colonization, we initially observe strong dissimilarity among 491 phyllosphere communities, regardless of their spatial position. During the summer, optimal 492 environmental conditions allow the progressive establishment of a cooperative and structured 493 bacterial community with a high yield strategy (51). At the end of the growing season, with 494 migration, environmental conditions shifting and leaves senescing, isolates with a fast-growth 495 strategy are able to grow rapidly, dominating the phyllosphere community and leading to its 496 homogeneization before leaves fully senesce. 497 498 Our study illustrates that Methylobacterium is a complex group of divergent lineages with 499 different ecological strategies and distributions, reflecting long-term adaptation to highly 500 contrasted environments. Based upon a similar observation, some authors recently proposed to 501 reclassify Methylobacterium group B within a new genus (Methylorubrum) that they argue is 502 ecologically and evolutionarily distinct from other Methylobacterium clades (30). Although clade 503 B was well supported as a distinct clade in our analyses, our results suggest that it is in fact 504 embedded within clade A, which would render the genus Methylobacterium paraphyletic if clade 505 B is defined as a distinct genus (Figure S5), and group B was not particularly ecologically 506 distinct in comparison with other major clades (Figure 1). Our results emphasize the fact that 507 torough genomic investigations are needed to clarify the taxomonic status of Methylobacterium. 508 Beyond any taxonomic considerations, neither clade identity assessed by individual genetic

17 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

509 markers nor the tremendous ecological diversity among Methylobacterium clades can predict all 510 of the spatial and temporal variation in Methylobacterium diversity in nature. In order to define 511 the niches of Methylobacterium clades and to understand the metabolic mechanisms underlying 512 their contrasted life strategies, future characterization of their functions and genome structure will 513 be required using phylogenomic approaches. 514 515 In conclusion, we find that Methylobacterium adaptive responses to local environmental variation 516 in the phyllosphere are driven by both long-term inherited ecological strategies that differ among 517 major clades within the genus, as well by seasonal changes affecting habitat characteristics and 518 community structure in the phyllosphere habitat. Overall, our study combining sequencing- and 519 culture-based approaches provides novel insights into the factors driving fine-scale adaptation of 520 microbes to their habitats, and in the case of Methylobacterium our approach revealed the 521 particular importance of considering organismal life-history strategies to help understand the 522 small-scale diversity and dynamic of this ecologically important taxon. 523 524 525 526

18 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

527 Figures and Table 528 529 Figure 1 - Methylobacterium phylogeny and ecology. Most of Methylobacterium diversity is found 530 in association with plants, especially in the phyllosphere. Phylogenetic consensus tree (nodal 531 posterior probabilities indicated next to the branches) from rpoB complete nucleotide sequences 532 available for 153 Methylobacterium genomes and rooted on 32 Methylobacteriaceae outgroups 533 (Microvirga, Enterovirga; no shown; see Supplementary dataset 1a). For each genome, species 534 name, the anthropogenic origin (black squares) and/or environmental origin (color code on top 535 right) are indicated. Groups A, B, C adapted from Green et Ardley (30). Monophyletic clades 536 within group A (A1-A9) were defined using a 92% nucleotide pairwise similarity cut-off on the 537 rpoB complete sequence. 538 539 Figure 2 - Sampling design. a) Locations of the two sampled forests MSH (green) and SBL 540 (orange) in the province of Québec (Canada). b) Time line survey in each forest in 2018 (2-4 time 541 points available per tree). c-d). Detailed map of each forest and each plot within forests (squares; 542 6 to 10 trees were sampled per plot; see Supplementary dataset 1b). In MSH, plots H0 and L0 543 were sampled once in 2017 for a pilot survey. In SBL and MSH, plots 1-6 were sampled 4 times 544 in 2018. For each plot, tree localizations are indicated by point colored according to their 545 taxonomie (color code on bottom left): ABBA (Abies balsamea), ACRU (Acer rubrum). ACSA 546 (Acer saccharum), OSVI (Ostrya virginiana), QURU (Quercus rubra), FAGR (Fagus 547 grandifolia), ASPE (Acer Pennsylvanicum). Shades of grey indicate elevation (50 m elevation 548 scale) 549 550 Figure 3 - Tests for phylogenetic association of traits with culture-based estimation of 551 Methylobacterium diversity. a) Part of variance in Methylobacterium isolated diversity 552 explained by each trait and their interactions (PERMANOVA tests for association; 10,000 553 permutations; x-axis) in function of pairwise nucleotide similarity as a proxy for phylogenetic 554 depth (PS; y-axis; see Supplementary dataset 1i). PERMANOVA were conducted at different 555 depths within a consensus phylogenetic tree (nodes with less than 30% of support were collapsed; 556 legend on top right) drawn from partial rpoB nucleotide sequences of 187 isolates (pilot survey in 557 2017: n=20; timeline survey in 2018: n=167) and 188 Methylobacteriaceae reference sequences.

19 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

558 The four following traits and their interactions were tested (see Venn diagram on top left for 559 color code): forest of origin, host tree species, sampling date and temperature of isolation. Points 560 indicate significant part of variance (legend on the left). b) Test for node association (100,000 561 permutations) with forest of origin and temperature of isolation (color code on top) mapped on 562 the rpoB phylogeny (scaled on PS values). Frames in the tree indicate nodes significantly 563 associated with at least one factor (ANOVA; Bonferroni correction; p<0.001: “***”; 564 p<0.01:”**”; p<0.05:”*”). For each isolate (names in bold), colored boxes at the tip of the tree 565 indicate forest of origin and temperature of isolation. 566 567 Figure 4 - Short-scale spatial and temporal dynamics of Methylobacterium communities 568 assessed by rpoB barcoding. a) A principal component analysis (PCA) on 200 569 Methylobacterium ASVs relative abundance (Hellinger transformation, Bray-Curtis (BC) 570 dissimilarity) shows that 179 phyllosphere samples cluster according to forest of origin (MSH: 571 open triangles, SBL: full triangles) and date of sampling (detail showed only for MSH). The 572 significant association of 83 and 25 ASVs with forest of origin and/or sampling date, respectively 573 (ANOVA, Bonferroni correction; p<0.05; point size proportional to variance; legend on bottom 574 left) is shown (points colored according to clade assignation; legend on top right). b) Spatial and 575 c) temporal autocorrelation analyzes conducted in each forest separately. Points represent BC 576 dissimilarity in function of pairwise geographic (pDist; b) or pairwise time (pTime; c) distance 577 separating two communities. For each forest and variable, the predicted linear regression 578 (BC∼pDIST or ∼pTime) is indicated (full line: p<0.001; dotted line: p>0.05; ANOVA). d) BC in 579 function of sampling time for each forest. e) Detail of spatial autocorrelation analyzes in MSH, 580 conducted for each sampling time point separately. f) Scaled ML tree (original tree from Figure 581 S1b scaled on pairwise nucleotide similarity (PS)) of 200 Methylobacterium ASV (points) and 582 176 reference rpoB sequences rooted on Microvirga and Enterovirga (out group not shown). 583 Temporal autocorrelation analyzes per forest were conducted for each node supported by at least 584 30% of bootstraps (200 permutations). For each forest and node, the strength of the slope 585 (estimate of the linear regression BC∼pTime; proportional to point size) was displayed when 586 highly significant (p<0.001; ANOVA, Bonferroni correction). 587

20 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

588 Figure 5 - Analysis of 79 Methylobacterium isolate growth performances under 4 different 589 temperature treatments. For each isolate and temperature treatment, the yield (Y: maximal 590 growth intensity) and growth rate (r; inverse of log+lag time) were estimated from growth 591 curves. a) Average growth curves (growth intensity in function of time) for each clade (line: 592 mean value; frame: 1/3 of standard deviation; point: average maximal growth). b) r in function of 593 Y. Each point represents the average r/Y values for an isolate and a temperature treatment (79 594 isolates x 4 treatments), colored according to clade membership. Ellipsoides are centered on 595 average values per clade and represent 30% of confidence interval (standard deviation). c) r (log 596 scale) in function of time at which samples strains were isolated from were collected, colored 597 according to the forest of origin. Points: real data; bars: average r value per forest (n=2) and time 598 (n=4) category. d) r in function of Y, corrected for clade assignement (residuals of the r~Clade 599 and Y~Clade linear regressions). Each point represents the average r/Y residual values for an 600 isolate and a temperature treatment (79 isolates x 4 treatments), colored according to monitoring 601 temperature (legend on top right). 602 603

21 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

604 Table 1 - PERMANOVA analysis of variance in Bacteria and Methylobacterium community 605 diversity. Bacteria diversity was assessed by 16s barcoding in 46 phyllosphere samples. 606 Methylobacterium was assessed by rpoB barcoding after filtering out non-Methylobacterium 607 diversity in 179 phyllopshere samples. Part of variance in dissimilarity (R2; Bray-Curtis index) 608 among samples associated with four factors and their possible interactions (F: forest of origin; D: 609 date of sampling; H: host tree species; P: plot within forest) and their significance are shown 610 (10,000 permutations on ASV relative abundance, Hellinger transformation; “***”: p<0.00l; 611 “**”: p<0.01; “*”: p<0.05.) For 16s, P was omitted to conserve degrees of freedom. Bacteria (16s) Methylobacterium (rpoB) Samples 46 179 Factor R2 Pr(>F) R2 Pr(>F) Forest of origin (F) 0.316*** <0.000 0.324*** <0.001 Host tree specie (H) 0.156*** <0.001 0.071*** <0.001 Time of sampling (D) 0.120* 0.016 0.048*** <0.001 Plot within forests (P) - - 0.080*** <0.001 F:H 0.020 0.080 0.004 0.110 H:D 0.239 0.217 0.074** 0.028 H:P - - 0.043** 0.007 D:P - - 0.058 0.455 H:D:P - - 0.085 0.052 Residuals 0.150 - 0.213 - 612 613 614

22 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

615 Table 2 - Summary of statistics from autocorrelation analyzes on 179 phyllosphere 616 Methylobacterium samples assessed by rpoB barcoding (200 ASVs). For each model, pairwise 617 dissimilarity between two communities was assessed with the Bray-Curtis dissimilarity index 618 (BC) from ASV relative abundance (Hellinger transformation) under a linear model. Spatial 619 autocorrelation general models : BC in function of pairwise spatial distance separating two 620 sampled trees (pDist) and date of sampling (Date) and their interaction (pDist:Date). Samples 621 from forests MSH and SBL were analyzed separately (two models). Only pairwise comparisons 622 among samples from a same date were considered. Spatial autocorrelation models per date: BC 623 in function of pairwise spatial distance (pDist). Each sampling date (n=4) and forest (n=2) was 624 analyzed separately (eight models). Temporal autocorrelation: general models: BC in function of 625 pairwise spatial time separating two sampled trees (pTime). Samples from forests MSH and SBL 626 were analyzed separately (two models) and all spatial scales were considered. For each model, 627 the average and standard deviation of the intercept (mean BC value) are indicated. For each 628 factor (pDist, Date, pDist:Date and pTime), the average and standard deviation of estimates 629 (slope) are indicated. Significance of estimates was assessed by ANOVA (“***”: p<0.00l; “**”: 630 p<0.01; “*”: p<0.05).

Categories (n) Intercept (sd) Estimates*10-3 (sd) Spatial autocorrelation general models: lm(BC∼pDist*D) Site (within dates) BC pDist D pDist:Date MSH 0.5965 (0.0107) -0.0041 (0.0192)*** -2.7648 (0.1313)*** 0.0007 (0.0002)** SBL 0.6493 (0.0097) 0.0157 (0.0145) -1.5575 (0.1646)*** 0.0000 (0.0002) Spatial autocorrelation models per date: lm(BC∼pDist) Site Date BC pDist MSH 27 Jun. 0.6237 (0.0340) -0.0425 (0.0725) 6 Aug. 0.4919 (0.0112) 0.0503 (0.0192)** 7 Sept. 0.3746 (0.0059) 0.0313 (0.0099)** 18 Oct. 0.2966 (0.0045) 0.0795 (0.0073)*** SBL 20 Jun. 0.6868 (0.0146) 0.0082 (0.0216) 16 Jul. 0.5819 (0.0113) 0.0215 (0.0174) 16 Aug. 0.5415 (0.0105) 0.0114 (0.0150) 20 Sept. 0.5222 (0.0089) 0.0145 (0.0130) Temporal autocorrelation general models (BC∼pTime) Site BC pTime MSH 0.4086 (0.0032) 1.0786 (0.0607)*** SBL 0.5789 (0.0030) 0.3012 (0.0617)*** 631

23 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

632 Table 3 - Part of variance in yield (Y) and growth rate (r) measured in 79 Methylobacterium 633 isolates grown under 4 temperature treatments. Y and r values were transformed in log to 634 meet normal distribution. Significance of Y and r response were evaluated by ANOVA in linear

635 models: log(Y)~F*H*D*Tp*TM*C and log(r)~ F*H*D*Tp*TM*C, respectively, with following 636 factors : clade (C), forest of origin (F), host tree species (H), time of sampling (D), temperature of

637 incubation during pre-conditioning (TP) and monitoring (TM) steps and their interactions (only 638 significant are shown: “***”: p<0.00l; “**”: p<0.01; “*”: p<0.05). 639 Rate Yield Forest (F) 0.015* 0.002 Host tree species (H) 0.022** 0.013** Date of sampling (D) 0.054*** 0.013** Pre-conditioning temperature (TP) 0.001 0.014** Monitoring temperature (TM) 0.158*** 0.020** Clade (C) 0.076*** 0.306*** F:D 0.006 0.036*** H:D 0.001 0.015** H:C 0.006 0.058*** D:C 0.032* 0.022* F:H:D 0.019** 0.035*** F:H:C 0.013* 0.006 other interactions 0.142 0.084 Residuals 0.456 0.377 640

641

24 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

642 References 643 1. Vorholt JA. 2012. Microbial life in the phyllosphere. 12. Nature Reviews Microbiology 644 10:828–840. 645 2. Lindow SE, Brandl MT. 2003. Microbiology of the Phyllosphere. Appl Environ 646 Microbiol 69:1875–1883. 647 3. Fürnkranz M, Wanek W, Richter A, Abell G, Rasche F, Sessitsch A. 2008. Nitrogen 648 fixation by phyllosphere bacteria associated with higher plants and their colonizing epiphytes of a 649 tropical lowland rainforest of Costa Rica. 5. The ISME Journal 2:561–570. 650 4. Laforest-Lapointe I, Messier C, Kembel SW. 2016. Host species identity, site and time 651 drive temperate tree phyllosphere bacterial community structure. Microbiome 4:27. 652 5. Laforest-Lapointe I, Messier C, Kembel SW. 2016. Tree phyllosphere bacterial 653 communities: exploring the magnitude of intra- and inter-individual variation among host 654 species. PeerJ 4:e2367. 655 6. Noble AS, Noe S, Clearwater MJ, Lee CK. 2020. A core phyllosphere microbiome exists 656 across distant populations of a tree species indigenous to New Zealand. PLOS ONE 657 15:e0237079. 658 7. Kembel SW, O’Connor TK, Arnold HK, Hubbell SP, Wright SJ, Green JL. 2014. 659 Relationships between phyllosphere bacterial communities and plant functional traits in a 660 neotropical forest. PNAS 111:13715–13720. 661 8. Shade A, Gregory Caporaso J, Handelsman J, Knight R, Fierer N. 2013. A meta-analysis 662 of changes in bacterial and archaeal communities with time. 8. The ISME Journal 7:1493–1506. 663 9. Malik AA, Martiny JBH, Brodie EL, Martiny AC, Treseder KK, Allison SD. 2020. 664 Defining trait-based microbial strategies with consequences for soil carbon cycling under climate 665 change. 1. The ISME Journal 14:1–9. 666 10. Moyes AB, Kueppers LM, PettRidge J, Carper DL, Vandehey N, O’Neil J, Frank AC. 667 2016. Evidence for foliar endophytic nitrogen fixation in a widely distributed subalpine conifer. 668 New Phytologist 210:657–668. 669 11. Lajoie G, Kembel SW. 2019. Making the Most of Trait-Based Approaches for Microbial 670 Ecology. Trends in Microbiology 27:814–823. 671 12. Nemergut DR, Schmidt SK, Fukami T, O’Neill SP, Bilinski TM, Stanish LF, Knelman 672 JE, Darcy JL, Lynch RC, Wickey P, Ferrenberg S. 2013. Patterns and Processes of Microbial

25 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

673 Community Assembly. Microbiol Mol Biol Rev 77:342–356. 674 13. CavenderBares J, Kozak KH, Fine PVA, Kembel SW. 2009. The merging of 675 community ecology and phylogenetic biology. Ecology Letters 12:693–715. 676 14. Martiny JBH, Jones SE, Lennon JT, Martiny AC. 2015. Microbiomes in light of traits: A 677 phylogenetic perspective. Science 350. 678 15. Tromas N, Taranu ZE, Castelli M, Pimentel JSM, Pereira DA, Marcoz R, Shapiro BJ, 679 Giani A. 2020. The evolution of realized niches within freshwater Synechococcus. 680 Environmental Microbiology 22:1238–1250. 681 16. Tromas N, Taranu ZE, Martin BD, Willis A, Fortin N, Greer CW, Shapiro BJ. 2018. 682 Niche Separation Increases With Genetic Distance Among Bloom-Forming Cyanobacteria. Front 683 Microbiol 9. 684 17. Thompson LR, Sanders JG, McDonald D, Amir A, Ladau J, Locey KJ, Prill RJ, Tripathi 685 A, Gibbons SM, Ackermann G, Navas-Molina JA, Janssen S, Kopylova E, Vázquez-Baeza Y, 686 González A, Morton JT, Mirarab S, Zech Xu Z, Jiang L, Haroon MF, Kanbar J, Zhu Q, Jin Song 687 S, Kosciolek T, Bokulich NA, Lefler J, Brislawn CJ, Humphrey G, Owens SM, Hampton- 688 Marcell J, Berg-Lyons D, McKenzie V, Fierer N, Fuhrman JA, Clauset A, Stevens RL, Shade A, 689 Pollard KS, Goodwin KD, Jansson JK, Gilbert JA, Knight R. 2017. A communal catalogue 690 reveals Earth’s multiscale microbial diversity. 7681. Nature 551:457–463. 691 18. Poretsky R, Rodriguez-R LM, Luo C, Tsementzi D, Konstantinidis KT. 2014. Strengths 692 and Limitations of 16S rRNA Gene Amplicon Sequencing in Revealing Temporal Microbial 693 Community Dynamics. PLOS ONE 9:e93827. 694 19. Ranjan R, Rani A, Metwally A, McGee HS, Perkins DL. 2016. Analysis of the 695 microbiome: Advantages of whole genome shotgun versus 16S amplicon sequencing. 696 Biochemical and Biophysical Research Communications 469:967–977. 697 20. Corpe WA, Rheem S. 1989. Ecology of the methylotrophic bacteria on living leaf 698 surfaces. FEMS Microbiol Ecol 5:243–249. 699 21. Keppler F, Hamilton JTG, Braß M, Röckmann T. 2006. Methane emissions from 700 terrestrial plants under aerobic conditions. 7073. Nature 439:187–191. 701 22. Clarke PH. 1983. The biochemistry of Methylotrophs by C. Anthony Academic Press; 702 London, New York, 1982 xvi + 432 pages. £24.00, $49.50. FEBS Letters 160:303–303. 703 23. Anthony C. 1991. Assimilation of carbon by methylotrophs. Biotechnology 18:79–109.

26 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

704 24. Ivanova EG, Doronina NV, Trotsenko YuA. 2001. Aerobic Methylobacteria Are Capable 705 of Synthesizing Auxins. Microbiology 70:392–397. 706 25. Madhaiyan M, Poonguzhali S, Lee HS, Hari K, Sundaram SP, Sa TM. 2005. Pink- 707 pigmented facultative methylotrophic bacteria accelerate germination, growth and yield of 708 sugarcane clone Co86032 (Saccharum officinarum L.). Biol Fertil Soils 41:350–358. 709 26. Madhaiyan M, Poonguzhali S, Sa T. 2007. Metal tolerating methylotrophic bacteria 710 reduces nickel and cadmium toxicity and promotes plant growth of tomato (Lycopersicon 711 esculentum L.). Chemosphere 69:220–228. 712 27. Dourado MN, Aparecida Camargo Neves A, Santos DS, Araújo WL. 2015. 713 Biotechnological and Agronomic Potential of Endophytic Pink-Pigmented Methylotrophic 714 Methylobacterium spp. Biomed Res Int 2015. 715 28. Ryu JH (Chungbuk NU, Madhaiyan M (Chungbuk NU, Poonguzhali S (Chungbuk NU, 716 Yim WJ (Chungbuk NU, Indiragandhi P (Chungbuk NU, Kim KA (Chungbuk NU, Anandham R 717 (Chungbuk NU, Yun JC (National I of AS and T, Kim KH (The U of S, Sa TM (Chungbuk NU. 718 2006. Plant Growth Substances Produced by Methylobacterium spp. and Their Effect on Tomato 719 (Lycopersicon esculentum L.) and Red Pepper (Capsicum annuum L.) Growth. Journal of 720 Microbiology and Biotechnology. 721 29. Lee HS, Madhaiyan M, Kim CW, Choi SJ, Chung KY, Sa TM. 2006. Physiological 722 enhancement of early growth of rice seedlings (Oryza sativa L.) by production of phytohormone 723 of N2-fixing methylotrophic isolates. Biol Fertil Soils 42:402–408. 724 30. Green PN, Ardley JK. 2018. Review of the genus Methylobacterium and closely related 725 organisms: a proposal that some Methylobacterium species be reclassified into a new genus, 726 Methylorubrum gen. nov. International Journal of Systematic and Evolutionary Microbiology 727 68:2727–2748. 728 31. Chen W-M, Cai C-Y, Li Z-H, Young C-C, Sheu S-Y. 2019. Methylobacterium 729 oryzihabitans sp. nov., isolated from water sampled from a rice paddy field. International Journal 730 of Systematic and Evolutionary Microbiology, 69:3843–3850. 731 32. Feng G-D, Chen W, Zhang X-J, Zhang J, Wang S-N, Zhu H. 2020. Methylobacterium 732 nonmethylotrophicum sp. nov., isolated from tungsten mine tailing. International Journal of 733 Systematic and Evolutionary Microbiology, 70:2867–2872. 734 33. Jia L juan, Zhang K shuai, Tang K, Meng J yu, Zheng C, Feng F ying. 2020.

27 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

735 Methylobacterium crusticola sp. nov., isolated from biological soil crusts. International Journal of 736 Systematic and Evolutionary Microbiology, 70:2089–2095. 737 34. Kim J, Chhetri G, Kim I, Lee B, Jang W, Kim MK, Seo T. 2020. Methylobacterium 738 terricola sp. nov., a gamma radiation-resistant bacterium isolated from gamma ray-irradiated soil. 739 International Journal of Systematic and Evolutionary Microbiology, 70:2449–2456. 740 35. Kim J, Chhetri G, Kim I, Kim MK, Seo T. 2020. Methylobacterium durans sp. nov., a 741 radiation-resistant bacterium isolated from gamma ray-irradiated soil. Antonie van Leeuwenhoek 742 113:211–220. 743 36. Ten LN, Li W, Elderiny NS, Kim MK, Lee S-Y, Rooney AP, Jung H-Y. 2020. 744 Methylobacterium segetis sp. nov., a novel member of the family Methylobacteriaceae isolated 745 from soil on Jeju Island. Arch Microbiol 202:747–754. 746 37. Jiang L, An D, Wang X, Zhang K, Li G, Lang L, Wang L, Jiang C, Jiang Y. 2020. 747 Methylobacterium planium sp. nov., isolated from a lichen sample. Arch Microbiol 202:1709– 748 1715. 749 38. Pascual JA, Ros M, Martínez J, Carmona F, Bernabé A, Torres R, Lucena T, Aznar R, 750 Arahal DR, Fernández F. 2020. Methylobacterium symbioticum sp. nov., a new species isolated 751 from spores of Glomus iranicum var. tenuihypharum. Curr Microbiol 77:2031–2041. 752 39. Marx CJ, Bringel F, Chistoserdova L, Moulin L, Farhan Ul Haque M, Fleischman DE, 753 Gruffaz C, Jourand P, Knief C, Lee M-C, Muller EEL, Nadalig T, Peyraud R, Roselli S, Russ L, 754 Goodwin LA, Ivanova N, Kyrpides N, Lajus A, Land ML, Médigue C, Mikhailova N, Nolan M, 755 Woyke T, Stolyar S, Vorholt JA, Vuilleumier S. 2012. Complete Genome Sequences of Six 756 Strains of the Genus Methylobacterium. J Bacteriol 194:4746–4748. 757 40. Tani A, Ogura Y, Hayashi T, Kimbara K. 2015. Complete Genome Sequence of 758 Methylobacterium aquaticum Strain 22A, Isolated from Racomitrium japonicum Moss. Genome 759 Announc 3. 760 41. Minami T, Ohtsubo Y, Anda M, Nagata Y, Tsuda M, Mitsui H, Sugawara M, 761 Minamisawa K. 2016. Complete Genome Sequence of Methylobacterium sp. Strain AMS5, an 762 Isolate from a Soybean Stem. Genome Announc 4. 763 42. Morohoshi T, Ikeda T. 2016. Complete Genome Sequence of Methylobacterium populi P- 764 1M, Isolated from Pink-Pigmented Household Biofilm. Genome Announc 4. 765 43. Belkhelfa S, Labadie K, Cruaud C, Aury J-M, Roche D, Bouzon M, Salanoubat M,

28 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

766 Döring V. 2018. Complete Genome Sequence of the Facultative Methylotroph Methylobacterium 767 extorquens TK 0001 Isolated from Soil in Poland. Genome Announc 6. 768 44. Vos M, Quince C, Pijl AS, Hollander M de, Kowalchuk GA. 2012. A Comparison of 769 rpoB and 16S rRNA as Markers in Pyrosequencing Studies of Bacterial Diversity. PLOS ONE 770 7:e30600. 771 45. Ogier J-C, Pagès S, Galan M, Barret M, Gaudriault S. 2019. rpoB, a promising marker for 772 analyzing the diversity of bacterial communities by amplicon sequencing. BMC Microbiol 773 19:171. 774 46. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. 2016. 775 DADA2: High resolution sample inference from Illumina amplicon data. Nat Methods 13:581– 776 583. 777 47. Green PN. 2015. Methylobacterium, p. 1–8. In Bergey’s Manual of Systematics of 778 Archaea and Bacteria. American Cancer Society. 779 48. Charron G, Leducq JB, Bertin C, Dube AK, Landry CR. 2014. Exploring the northern 780 limit of the distribution of Saccharomyces cerevisiae and Saccharomyces paradoxus in North 781 America. FEMS Yeast Res 14:281–8. 782 49. Leducq JB, Charron G, Samani P, Dube AK, Sylvester K, James B, Almeida P, Sampaio 783 JP, Hittinger CT, Bell G, Landry CR. 2014. Local climatic adaptation in a widespread 784 microorganism. Proceedings Biological sciences / The Royal Society 281:20132472. 785 50. Zwietering MH, Jongenburger I, Rombouts FM, Riet K van ’t. 1990. Modeling of the 786 Bacterial Growth Curve. Appl Environ Microbiol 56:1875–1881. 787 51. Lipson DA. 2015. The complex relationship between microbial growth rate and yield and 788 its implications for ecosystem processes. Front Microbiol 6. 789 52. MacArthur RH, Wilson EO. 2001. The Theory of Island Biogeography. Princeton 790 University Press. 791 53. CavenderBares J, Ackerly DD, Baum DA, Bazzaz FA. 2004. Phylogenetic 792 Overdispersion in Floridian Oak Communities. The American Naturalist 163:823–843. 793

29 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Environmental sources Plant (unknown part) Phyllosphere Endophyte Seed Rhizosphere C Soil Fungi or lichen M. radiotolerans

M. indicum Water UNC300MFChir4.1 M. platani Ocean M. organophilum

285MFTsu5.1

13MFTsu3.1M2 275MFSha3.1 Gut M. aquaticum Anthopogenic

M. oryzae 190mf

yr668 BK227 A9 B34 unknown Leaf361 NBRC15690 M. tarhaniae JCM2831 NS228

NS229 RE1.2 NS230 SE2.11 SE3.6 ES_PA−B5 C1 ap11 MA−22A M. fujisawaense MAMP4754 yr596 M. currus ME94 17Sr1−28 PMB02 M. variabile JCM14648 DSM760 DB0501 CBMB20 17Sr1−39 YL−MPn5−2016 174MFSha1.1 M. crusticola 17Sr1−1 XJLW 100 YL−MPn6−2016 100 DSM25844 M. nodulans M. phyllosphaerae 100 100 100 DSM16371 100 78 100 B1 98 100 100 PR1016A 100 DSM16961 M. mesophilicum DSM5686 UNCCL125 79 100 100 6HR−1 A5 100 100 MIMD6 UNCCL110 100 100 87 100 100 ORS2060 M. segetis M. phyllostachyos CBMB27 80 4−46 M. durans 100 WSM2598 CBMB27 100 100 TER−1 M. oxalidis 100 SR1.6/6 98 17J42−1 M. soli 100 100 100 100 100 100 17SD2−17 90 100 2A 97 100 95 100 NBRC107715 UNC378MFBL47 100 100 YIM48816 100 100 M. brachiatum P1−11 YIM132548 100 TX0642 90 100 100 Leaf465 76 100 Leaf89 GXF4 111MFTsu3.1M4 90 86 Leaf88 93 100 M. pseudosasicola Leaf94 ARG−1 100 100 100 Leaf111 100 BL36 92 Leaf104 100 GXS13 94 V23 100 100 WL7 99 Leaf125 97 80 87 WL18 100 GV094 100 100 100 WL64 100 GV104 100 99 100 100 100 100 Leaf117 WL1 100 WL2 99 Leaf113 100 A1 81 BTF04 17Sr1−43 96 100 100 100 Gh−105 CCH5−D2 84 WL69 M. gossipiicola SW08−7 100 100 Leaf99 100 100 100 100 A7 SB0023/3 100 Leaf100 100 WL120 100 M. dankookense 100 Leaf87 WL8 100 100 Leaf112 M. symbioticum WL12 100 100 100 Leaf469 A8 100 WL103 100 Leaf102 WL6 100 100 77 100 100 100 Leaf86 WL116 WL93 100 Leaf91 87 100 100 100 10 WL30 100 94 100 100 100 88A WL119 Leaf106 100 100 97 100 Leaf85 100 DSM2163 100 Leaf93 100 Leaf456 WL19 100 100 100 L1A1 75 Leaf399 85 A6 100 100 Leaf108 100 100 CD11_7 Leaf466 CGMCC1.6474 BJ001 NBRC107716 DSM24105 WL9 NBRC107714 Q1 M. rhodinum R2−1 YC−XJ1 AMS5 DSM11490 Leaf123 P−1M CP3 DM4 DSM13060 PinkelPinkel_01 PA1 AM1 DM4

RAS18 PSBB040 DM1 M. salsuginis CLZ B4 DSM5687DB1607 MB200 NI91

CM4 M. gnaphalii A2

TK0001 M. brachythecii

PSBB041

M. populi M. haplocladii NBRC15911

M. extorquens A3 M. rhodesianum A4

B bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. a. b. 9-jui 9-jui CANADA ON a b SBL 1

29-jui 29-jui June 2 CANADA 4 19-jui 19-jui 3 QC July m 500

QC 8-aoû N USA

NFL Aug. 5 28-aoû 28-aoû MSH MSH 4 3 NB NB SBL Atlantic H0 PEI

Ocean 2 NB 17-sep Sep. 200 m 200

MSH 1 L0 PEI NS NS 6

ON NS 200 km N

USA 7-oct Oct. 200 km 200 27-oct 27-oct

c d Atlantic Ocean

SBL MSH NFL

300 MSH mSBL 300 m 03 3 m 05 H0

04 04 L0 03 02 02 3 m

01

01 06

3 m ABBA ACRU ACSA OSVI QURU FAGR 3 m 3 m 3 m 3 m 3 m 3 m 3 m ACPE bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a b B

Nodal support Forest of origin SBL MSH A5a 30 65 100 Host tree Sampling Isolation temperature 30°C 20°C

M. populi M. populi M. thiocyanatum M. populi M. populi M. populi

M. extorquens

species time M. extorquens M. extorquens M. extorquens

M. extorquens M. zatmanii M. extorquens M. extorquens M. extorquens Forest of Isolation PS M. extorquens

1 XJ1 M. rhodesianum

− 048 (2018) 1M − 2 − − 1 043 (2018) temperature − M. salsuginis 014 (2018) − E R2 P − PinkelPinkel_01 037 (2018) DSM11490 YC BJ001 M. rhodinum CD11_7 origin B4 − J − Leaf123 CP3 J J AM1 M. oxalidis DSM13060 DM4 DM4 RAS18 PA1 PSBB041 064 (2018) TK0001 003 (2018) NBRC15911 − CM4 − 1 E E M. brachiatum Q1 103 (2018) AMS5 − M. brachiatum PSBB040 J M. pseudosasicola DM1 NI91 CLZ DB1607 DSM5687 17 MB200 − L1A1 1 CGMCC1.6474 − DSM2163 1 Leaf456 − NBRC107715 17SD2 17J42 0.99 ARG GXF4 111MFTsu3.1M4 TX0642 066 (2018) BL36− 085 (2018) J − J LYS072 (2017)135 (2018) LYS080− (2017) J DNA012 (2017) 133 (2018) DNA011− (2017) J 074 (2018) − 141 (2018) J − J 024 (2018) 0.98 − 104 (2018) E − 138 (2018) E − J 008 (2018) − 002 (2018) E − J 121 (2018) − 107 (2018) E − E 001 (2018) − 002 (2018) J − p 0.001 E 026 (2018) ≤ − E 107 (2018) J− 009 (2018) 0.97 J− GXS13−008 (2018) p ≤ 0.01 J 121 (2018) − J −011 (2018) E 077 (2018) J− −028 (2018) J 010 (2018) p ≤ 0.05 J− −075 (2018) J 130 (2018) E− −115 (2018) E 116 (2018) 0.96 J− −033 (2018) E 028 (2018) E− −062 (2018) E 051 (2018) J− −018 (2018) J 072 (2018) J− −021 (2018) J 123 (2018) J− −102 (2018) 0.95 E 148 (2018) J− −066 (2018) E 129 (2018) E− −128 (2018) J 047 (2018) E− J−033 (2018) 0.11 0.08 0.06 0.03 0 J−017 (2018) J−019 (2018) J−087 (2018) J−086 (2018) J−048 (2018) J−052 (2018) E−010 (2018) J−113 (2018) E−128 (2018) E−027 (2018) E−022 (2018) WL64 WL1 WL2 Part of variance J−111 (2018) WL18 WL7 E−035 (2018) J−118 (2018) A9 J−026 (2018) E−034 (2018) J−124 (2018) J−147 (2018) J−115 (2018) E−012 (2018) E−125 (2018) SBL *** J−036 (2018) SBL ** J−011 (2018) DSM25903 E−124 (2018) DB1703 E−101 (2018) Enterovirga 17mud1−3 E−029 (2018) HR1 30°C*** J−120 (2018) R24825 J−070 (2018) DSM14364 E−021 (2018) DSM21344 SR1.6/6 XJLW M. mesophilicum NBRC106136 CDVBN77 B1 c23x22 CBMB20 −1258 DSM5686 M. oryzae NCCP YL −MPn6−2016 M. fujisawaense KCTC23863 YL−MPn5 BR3299 CBMB27 −2016 KLBC81 CBMB27 −2 UNCCL110−1 M. phyllosphaerae CGMCC1.7666−Q2068 M. phyllosphaerae Marseille UNCCL125 CCBAU65841 Leaf361 WSM3557 yr668 R24845 817 UNC300MFChir4.1 − BK227 ATCCBAA DSM15743 190mf BSC39 285MFTsu5.1 AT3.9 M8 B34 Microvirga 13MFTsu3.1M2 Lmie10 275MFSha3.1 V5/3m ME94 DSM760 ES_PA SYSUG3D203c27j1 SYSUG3D207−2 MAMP4754−B5 JCM2831 M. radiotolerans JC119 M. organophilum 023101 C1 − − M. radiotolerans 46 RE1.2 −HGUTJC1194− NBRC15690 M. radiotolerans J MGYG −164 (2018) M. radiotolerans WSM2598 2A ORS2060 P1 DB0501 UNC378MF−11 M. radiotolerans BL47 M. radiotolerans DSM25844 E DSM16371 −005 (2018) PR1016A 1 J− − J 104 (2018) M. nodulans 39 −105 (2018) 17Sr1− J− 174MFSha1.1 E 158 (2018) 17Sr1 −119 (2018) M. phyllostachyos M. tarhaniae PMB02 E− J 114 (2018) M. aquaticum −003 (2018) M. currus JCM14648SE2.11 J− SE3.6 J 106 (2018) −132 (2018) NS228 E− NS230 J− 009 (2018) NS229 J 064 (2018) −22A −082 (2018) M. platani J− MA yr596 E 127 (2018) M. platani ap11 − −28 E 127 (2018) M. platani J −109 (2018) M. platani E −057 (2018) 17Sr1 1 J − − 1 − 065 (2018) M. indicum DSM16961MIMD6 − J 045 (2018) M. indicum 6HR J −088 (2018) TER J − M. indicum − 049 (2018) E 122 (2018) J − − 126 (2018) M. aquaticum WL9 E 093 (2018) D2 − NBRC107716DSM24105 − 43 E 046 (2018) − 7 E − − − 116 (2018) NBRC107714 J 006 (2018) CCH5 J − M. variabile 17Sr1 − 030 (2018) SW08 DNA020083 (2017)(2018) E C SB0023/3 J − − 042 (2018) DNA021076 (2017)(2018) YIM132548 E YIM48816Leaf91 77 J − M. gnaphalii 123 (2018)Leaf86 J − 061 (2018) − E − 067 (2018) E 054 (2018) M. brachythecii Leaf93 J − M. haplocladii E − 030 (2018) J − 015 (2018) J − 103 (2018) 047 (2018) WL19 J − 150 (2018) − 88A J − 065 (2018) J Leaf85 J − 163 (2018) 10 E − 062 (2018) M. soli LYS093 (2017) J − 022 (2018) LYS069 (2017) J − 039 (2018) M. dankookense Leaf106 E − 059 (2018) 112 (2018) J − 136 (2018) − J − 063 (2018) J E − 073 (2018) J − 005 (2018) 045 (2018) Leaf89 J − LYS037− (2017) 106 (2018) − 078 (2018) Leaf88 E 160 (2018) E − 068 (2018) Leaf94 J 029 (2018) − J − J − Leaf111 J 040 (2018) J E − 134 (2018) Leaf465GV104 J − 117 (2018) GV094 E − 137 (2018) J − J − 120 (2018) E − 152 (2018) E − 020 (2018) J − 092 (2018) J 156 (2018) DNA007 (2017) − − 105 (2018) A4 J − V23 E 016 (2018) J 031 (2018) 132 (2018) J − 089 (2018) J − 108 (2018) WL93 − − WL116 − 146 (2018) WL119 − WL30 025 (2018) LYS051 (2017) − J 125 (2018) J WL6 130 (2018) E E E J 139 (2018) E DNA013 (2017) WL8 J WL103 WL120 − DNA024 (2017) −

− − − − 131 (2018) 050 (2018)

A7 142 (2018) 112 (2018) 041 (2018) DNA001 (2017) Leaf125 037 (2018) 084 (2018) Leaf104 DNA010LYS083 (2017) (2017) Leaf399 105 153 (2018) Leaf466 WL69 − 056 (2018) Leaf108 BTF04 Leaf113 − Leaf99 DNA006 (2017) − Leaf117 DNA014 (2017) J − WL12 A8 Leaf87 LYS027 (2017) J J Leaf112

Leaf102 Leaf469 Leaf100 Gh

090 (2018)

162 (2018) −

− J

J

DNA018 (2017) A5b A10 A2 M. gossipiicola

A3 A1 A6 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1.0 Forest of origin 1.0 Forest of origin a Clades b MSH c MSH SBL SBL A1 A4 A9 5 A2 A5 A10 A3 A6 B

18 Oct. 0.8 0.8

0 27 Jun. SBL ity (BC) ity (BC) MSH 7 Sep. 6 Aug. 0.6 0.6 −5 Axis 2 (4.58%) unity dissimilar unity dissimilar Part of var. / Factor −10 0.04 0.4 0.4 0.1 0.2 Comm Comm 0.4 Forest Date ASV contribution −15 & Forest/Date association (ANOVA) 0.2 0.2

−10 −5 0 5 10 0 2 8 44 234 1257 0 26 54 113

Axis 1 (15.38%) pDistance (log scale, m) pTime (log scale, days)

slope lm(BC~pTime) d Forest of origin e MSH f C 0.004 1.0 0.003 MSH 27 Jun. NS228 Meth−ASV−160

Meth−ASV−1228 Leaf361 NS229 275MFSha3.1 13MFTsu3.1M2 yr668 NS230 UNC300MFChir4.1 SE2.11 BK227 285MFTsu5.1 MA−22A B34 190mf ME94 0.002 SE3.6 MAMP4754 17Sr1−39 ES_PA−B5 DSM760 JCM14648 yr596 JCM2831 174MFSha1.1DSM16371 NBRC15690 C1 PMB02 ap11 RE1.2 UNCCL125 PR1016A UNCCL110 DSM5686 SBL 6 Aug. DSM25844 CBMB27−1 17Sr1−1 CBMB27−2 DSM16961 B1 CBMB20 XJLW 0.001 17Sr1−28DB0501 YL−MPn5−2016 YL−MPn6−2016 0.8 UNC378MF 6HR−1 P1−11 Meth−ASV−110 MIMD6 WSM2598 Meth−ASV−96 7 Sep. Meth−ASV−942 Meth−ASV−89 ORS2060 Meth−ASV−669 4−46 Meth−ASV−181 Meth−ASV−1286 Meth−ASV−1138 Meth−ASV−1079 TER−1 Meth−ASV−1139 Meth−ASV−83 Meth−ASV−104 Meth−ASV−40 A4 Meth−ASV−589 SR1.6/6 Meth−ASV−277 Meth−ASV−427 A9 18 Oct. Meth−ASV−115 Meth−ASV−710 Meth−ASV−98 Meth−ASV−506 Meth−ASV−151 Meth−ASV−908 Meth−ASV−469 Meth−ASV−1362 Meth−ASV−11 Meth−ASV−1380 Meth−ASV−82 Meth−ASV−13 Meth−ASV−529 Meth−ASV−1064 Meth−ASV−406 2A Meth−ASV−414 Meth−ASV−108 Meth−ASV−887 BL47 Meth−ASV−58 NBRC107716 Meth−ASV−1279 Meth−ASV−1305 Meth−ASV−785 DSM24105 Meth−ASV−1164 Meth−ASV−371 Meth−ASV−182 NBRC107714 Meth−ASV−453 0.8 0.7 Meth−ASV−1391 Meth−ASV−379 Meth−ASV−906 Meth−ASV−265 Meth−ASV−389 Meth−ASV−401 Meth−ASV−27 TK0001WL9 Meth−ASV−578 PSBB041 Meth−ASV−362 Meth−ASV−106 PA1 Meth−ASV−824 NBRC15911CM4 ity (BC) ity (BC) Meth−ASV−390 BL36 RAS18 WL7 DM4−1 Meth−ASV−601 DM4−2 Meth−ASV−777 Meth−ASV−287 AM1 WL64 DSM13060 WL2 Meth−ASV−474 Q1 WL1 Meth−ASV−833AMS5 Meth−ASV−694 Meth−ASV−382 WL18 111MFTsu3.1M4 0.6 PSBB040 TX0642 Meth−ASV−381CP3 GXF4 ARG−1 Leaf123 Meth−ASV−705 Meth−ASV−20 Meth−ASV−911B4 Meth−ASV−569 Meth−ASV−373 NI91 Meth−ASV−1154 CLZ Meth−ASV−657 DM1 DSM5687 Meth−ASV−1096 Meth−ASV−1316 DB1607 Meth−ASV−449 MB200 Meth−ASV−649 DSM11490 0.6 PinkelPinkel_01 Meth−ASV−1091 Meth−ASV−279 BJ001 Meth−ASV−388 CD11_7 Meth−ASV−242 YC−XJ1 Meth−ASV−1059 R2−1 Meth−ASV−299 0.5 B P−1M GXS13 Leaf456 Meth−ASV−673 DSM2163 Meth−ASV−2 L1A1 Meth−ASV−80 Meth−ASV−1267 DNA012 Meth−ASV−504 LYS080 DNA011 Meth−ASV−490 Meth−ASV−231 Meth−ASV−369 LYS072 Meth−ASV−140 Meth−ASV−1211 Meth−ASV−594 Meth−ASV−920 Meth−ASV−612 Meth−ASV−814 Meth−ASV−1115 Meth−ASV−1232 Meth−ASV−756 Meth−ASV−762 Meth−ASV−7 Meth−ASV−1235 unity dissimilar unity dissimilar Meth−ASV−123 Meth−ASV−521 WL8 Meth−ASV−1271 WL120 Meth−ASV−680 Meth−ASV−48 WL12 WL103 0.4 Meth−ASV−935 Meth−ASV−204 Meth−ASV−270 Meth−ASV−209 Meth−ASV−146 DNA013 Meth−ASV−125 Meth−ASV−316 Meth−ASV−691 Meth−ASV−294 Meth−ASV−509 Meth−ASV−178 LYS037 0.4 Meth−ASV−1121 Meth−ASV−445 Meth−ASV−582 Meth−ASV−29 Meth−ASV−932 Meth−ASV−187 Meth−ASV−212 Meth−ASV−333 Meth−ASV−1356 Meth−ASV−35 Meth−ASV−809 Meth−ASV−957 CGMCC1.647417J42−1 Meth−ASV−590 Meth−ASV−693 Meth−ASV−126 Meth−ASV−114917SD2−17

Comm Comm Meth−ASV−751 NBRC107715CCH5−D2 Meth−ASV−12 17Sr1−43 Meth−ASV−769 SB0023/3 Meth−ASV−1387 SW08−7 Meth−ASV−1253 0.3 Meth−ASV−256 Meth−ASV−424 Meth−ASV−997DNA021 Meth−ASV−6 Meth−ASV−821 Meth−ASV−968 Meth−ASV−473 A5a DNA020 Meth−ASV−168 Meth−ASV−167 WL6 A6 Meth−ASV−26 A7 Meth−ASV−78 Meth−ASV−1354 LYS083 Meth−ASV−479 Meth−ASV−74 DNA014 DNA007 Meth−ASV−421 A8 Meth−ASV−564 DNA006 WL30 Meth−ASV−21 WL93 Meth−ASV−77 WL119 WL116 Meth−ASV−1203 Meth−ASV−141 Meth−ASV−749 Meth−ASV−900 Meth−ASV−525 Leaf125 Meth−ASV−951 LYS027 Meth−ASV−915 Meth−ASV−577 DNA010 V23 Meth−ASV−1395 A10 Meth−ASV−191 YIM132548 DNA001 YIM48816 LYS051 10 Meth−ASV−796 Meth−ASV−1058 DNA024 Meth−ASV−1340 LYS093 Meth−ASV−733 LYS069 Meth−ASV−44 WL19 0.2 Leaf85 0.2 Leaf106 88A Meth−ASV−486Meth−ASV−50 Leaf93 Meth−ASV−882 77 Leaf91 Meth−ASV−574 Leaf86 Meth−ASV−133 Meth−ASV−904 Meth−ASV−1179 Meth−ASV−707 Meth−ASV−571Meth−ASV−67 Meth−ASV−1061 Leaf108 Leaf399 Leaf466 BTF04 Leaf113 Leaf117 Meth−ASV−482 Meth−ASV−666 Gh−105 Leaf100 Leaf102 Leaf469 Meth−ASV−1388Meth−ASV−709 Leaf112 Meth−ASV−291 Leaf104 Meth−ASV−25

Meth−ASV−1207 Leaf111 Leaf88 Leaf465 Leaf89 Meth−ASV−927 Leaf94 WL69 GV094 Meth−ASV−475 GV104 Meth−ASV−320 Leaf99 Meth−ASV−283 Leaf87 Meth−ASV−122 DNA018

Meth−ASV−697 Meth−ASV−1083

Meth−ASV−132 Meth−ASV−375 Meth−ASV−960 Meth−ASV−368 A5b Forest of origin June Aug. Sept. Oct. 0 3 12 56 259 1204 MSH A1 A2 A3 SBL Sampling date pDistance (log scale, m) bioRxiv preprint doi: https://doi.org/10.1101/2021.06.04.447128; this version posted July 7, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a b Clade Clade A1 A1 15 A10 20 A10 A2 A2 A6 A6 A9 15 A9 10 B B

ield (Y) 10 Y 5 Average growth Average 5

0 0

0 7 13 24 0.04 0.08 0.12 0.16

Time (days) Growth rate (r) c d 0.16 15 Monitoning temperature 0.14 20 °C 30 °C 0.12 10

0.10 5 0.08 ate (log scale) 0

wth r 0.06 Forest Residual lm(Y~clade) Gro SBL −5 MSH 0.04

June July Aug. Sept. Oct. −0.06 −0.02 0.02 0.06

Sampling date Residual lm(r~clade)